Good News from Washington

I just got back from the National Academy of Sciences meeting that was mandated in the 2003 Nano Act to study "molecular self-assembly." As far as I can tell, it went extremely well. I spoke in the second session, on definitions, but I was there for all six sessions. More details on that later; for tonight I'll just give initial impressions.

The committee focused directly on molecular manufacturing--which is good, because the wording in the near-final version of the law was "molecular manufacturing" before a nano business organization lobbied to change it to scuttle the study.

(In what follows, MM is molecular manufacturing; NNI is National Nanotechnology Initiative.)

There were some skeptics on the committee, and some who seemed more willing to accept that MM could work. But everyone there was clearly very smart, and everyone seemed intellectually honest. I'm pretty confident that MM will get a fair hearing--that they won't write a report which merely endorses the unscientific and off-topic objections that have been floating around for the last decade or two. Whatever they produce will be the result of open-minded and very careful study.

At the end of the second day, the committee went around the table with each person giving their personal reactions to what they'd seen so far--two sessions of history, one of definitions, and two of technology. A few members had to leave early, and a few said they were still absorbing what they'd heard. I took notes, but not quotes, and I was scribbling in a hurry, so I won't give people's names lest I mis-attribute an idea to them. But this is what I heard:

A thousand miles starts with a single step. We're busy proving A->B. We have no proof for E->F. (Where F is molecular manufacturing.)

There was impressive research in the last session. It somewhat alleviated my fears [about MM being impossible?]. We need a wide net for nanoscale tools--we don't know where is the tipping point for a breakthrough. Maybe we need a DARPA-like high risk/high reward pot to see what's possible.

I did biopolymers and thought it was nanotechnology; now I see we need definitions.

I've been through tech revolutions--lasers and computer chips. Big dreams are good; hats off to Drexler. However--it drives me crazy when people do computations without taking physical issues into account. If you say "manufacturing" you inherit physical issues--tolerances, yield, etc.

I work with nanotech companies--I see and appreciate proofs of concept--there are lots of [research?] directions. We need to survey the landscape widely to avoid being surprised by competitive innovations.

We're hearing that MM is a done deal--but we're being asked to consider if it's possible. I can't forget big picture/system. The devil is in the details. Why don't we have advanced e.g. medical products already?

The books Discoverers and Creators talk about history of technology. We owe a debt of discovery / another chapter in those books to Drexler; he's why we're sitting here. We're at a critical point: what will that chapter look like?

Thanks to the professors [in the tech sessions] for showing that the first steps are done. I advise patience to the advance thinkers. I did a heretical research. In 1972 I was told: show the first step. I did it. Then the engineers didn't care. So I showed them samples. *Today* my material is first being used in jet engines. (I had given up in '96.) We need demonstrations and patience.

I'm a scientist and engineer. Want to thank the panelists and visionaries. And Roco, for vision. Nanomanufacturing is in the NNI vision. [Drexler interrupts: nanomanufacturing is different from MM--the NNI is trying to muddy the waters.] I hope we can find a way to get materials into jet engines more rapidly. [Previous speaker interrupts--No! Be conservative with jet engines!] Let's se a grand challenge: an artificial ribosome. How stripped-down can we get a cell with a reasonable error rate for manufacturing? Keep a comparable goal in mind.

We're still in the process of identifying tasks. I did structure/function relationships of proteins and enzymes. Cells do have error correction--I'm not sure we're near achieving that. We need much more terminology work.

The NNI has come a long way. There's a lot going on.

Thanks everyone. I appreciated the civil tone of the discourse, and the enthusiasm. The passion for ideas was good to see. As someone else said: we haven't talked about: is MM desirable? At some point we need to look at societal needs we could address with this tech [something about public funds] - given fundamental research, keep in mind "to what end?" To the extent researchers can do that, it'll be helpful - resonant - public issues: health, energy.

I see progress toward MM: biology, chemistry, mechanical. We need to know more about fundamental limitations; I was impressed by a presentation on thermal noise.

I'm sure different people will see different things in the meeting, and these brief personal summaries don't even begin to give a full picture of the structure and content of the meeting. But I think they show the spirit of questioning that we need to begin a full and careful discussion of the technical merits of molecular manufacturing. I don't expect the committee to say either "Yes it's workable" or "no it's not." But "We looked at it for months and couldn't rule it out" would be a significant step forward.

Chris

Comments

[Mike Treder, CRN]

Great report, Chris, and thanks for all your hard work in preparing materials for the NAS committee meeting.

I'd be very interested in hearing your response to each of the points of view you reported above, if you think it's appropriate.

[Philip Moriarty]

Chris,

A very well-written and extremely interesting piece on the NAS committee meeting - thank you for posting this.

I've posted some comments re. definitions (and other comments put forward at the committee meeting) at:http://nanobot.blogspot.com/2005/02/molecular-manufacturing-back-on-table.html. Although some aspects of that post might be seen as 'retreading old ground', I'm concerned by comments at the NAS meeting that seem to suggest that it's difficult to define what's meant by (or what's required for) 'molecular manufacturing' in the Drexlerian sense. I'm of the opinion that, from a reading of "Nanosystems", it's not at all difficult to grasp the fundamental concept underlying molecular manufacturing. However, nebulous terminology makes it so much easier to 'move the goalposts'...

Best wishes,

Philip

[Chris Phoenix, CRN]

Philip, your URL didn't work. I look forward to reading your comments.

To make a clickable link in a comment here, you can put < a href="http://crnano.org">CRN's Website< /a> around the words you want to be clickable. Delete the spaces immediately following the < signs. (But your URL didn't work even when I cut-and-pasted it.)

Much of the difficulty in definition came from the substitution of "molecular self-assembly" for "molecular manufacturing" late in the lawmaking process. If that hadn't happened, the committee could have saved a lot of time by just agreeing to start from Drexler's definition--which was what was intended, because the Foresight Institute helped write the original language.

It's not easy to delineate exactly what concepts and approaches are associated with the field of molecular manufacturing. Does it require mechanical engineering at the nanoscale? No, but mechanical engineering looks like an efficient approach, a good fit for the mechanical approach to structure-building. So it's not a required concept, but it's close to the core.

Does it require machine-phase chemistry? No, certainly not. But mechanisms in vacuum may be more efficient due to less drag, and they're unfamiliar, so they're interesting to talk about. Note that there are gray areas even in the definition of machine-phase chemistry: does a system in which all reactive molecules are controlled near every reaction, but some transport is done by liquid-xenon solvent, count as machine-phase?

Does molecular manufacturing even require covalent chemistry in the placement step? How would we classify a system where molecular building blocks are placed non-covalently, then subsequently crosslinked in a bulk reaction (perhaps photon-caused)? The position of each reaction will have been selected programmably, but not at the time of the reaction.

To take another example, Drexler's definition in Nanosystems requires positional selection of reaction sites. Does Seeman's programmable DNA-building DNA machine count? Well, if you program the machine and then add the feedstock, then the reaction happens by self-assembly, as in enzymes; strictly speaking, this is not mechanosynthesis. But if you let the feedstock bind to the machine and then program it to rotate the molecules into position to react, then that looks more like mechanosynthesis. (In fact I think it works by the first method.)

Any human-language definition will have gray areas. I agree with you that the core concept of molecular manufacturing--the use of programmable mechanisms to control fabrication of molecular structures--isn't too hard to state and use.

Then there's the question of how useful the products are. If a Seeman-style machine could be made to build a complete duplicate of itself, that would be close to if not actually achieving molecular manufacturing. But it wouldn't come close to building the high-performance products associated with molecular manufacturing.

Of course, any field has gray areas. Metallurgy would seem to be easy to define--but what about cheap "pot metal" alloys that may contain plastic filler? What about diamond coatings on steel? Does welding count as metallurgy? Etc.

I'm all in favor of less nebulous terminology. Nebulous terminology and lack of specificity is what has allowed the NNI to think it's funding lots of work that's relevant to molecular manufacturing, when most of it is really not relevant. (I'm composing a blog post on this.) Molecular manufacturing people have a strong incentive to exclude things that are "nanotech" but do not lead to high-throughput positional fabrication.

If it looks like we're now trying to "move the goalposts" and include too much, that may be because the definition has always been that broad, but people have until recently been focusing on a very narrow proposal (i.e. diamond built in vacuum). Don't forget that protein engineering was the first proposed avenue toward molecular manufaturing--proposed in a PNAS paper in 1981 by Drexler.

Chris

[Philip Moriarty]

Chris,

Thanks for the advice on hyperlinks. I (rather stupidly) didn't realise that conventional HTML would work.

The link is:

Molecular manufacturing back on the table.

I'll respond to the detail of your comments when I get a chance ('neck deep' in teaching, research and admin at the moment).

For now, I note that you have recently posted an essay on molecular manufacturing on the CRN website. My key question is whether CRN still stands by the following definition of molecular manufacturing: "Molecular manufacturing (MM) means the ability to build devices, machines, and eventually whole products with every atom in its specified place" (as defined here:
Timeline for Molecular Manufacturing .)

Your statement of the core concept of molecular manufacturing in your post above ("the use of programmable mechanisms to control fabrication of molecular structures") is a much more 'nebuluous' description than the definition above. Moreover, it's much more vague than Drexler's definitions in Nanosystems:

“Mechanosynthesis: chemical synthesis controlled by mechanical systems operating with atomic-scale precision, enabling direct positional selection of reaction sites”

“Molecular manufacturing: The production of complex structures via nonbiological mechanosynthesis”.

The fundamental principle of molecular manufacturing, is the molecule-by-molecule (or atom-by-atom) construction of structures by *atomic precision* (as defined by Drexler) computer control. If molecular manufacturing is to be redefined so that it doesn't involve atomic precision, computer-controlled chemistry, then CRN is most definitely "moving the goalposts". (Seeman's work could possibly be seen as a **step towards** molecular manufacturing (although I'd be very interested in Seeman's opinions on the matter) - see Chapter 16 of Nanosystems - but it's very, very far away from an implementation of molecular manufacturing as defined in Nanosystems).

People have focussed on a very narrow proposal (diamond built in vacuum) because Drexler puts forward this "very narrow proposal" as his prototypical platform for mechanosynthesis (Chapter 8 of "Nanosystems"). Indeed, you'll find that some (perhaps many?) in the MNT community don't see diamondoid mechanochemistry as a "very narrow proposal" at all (see, for example, Making and doing ). Moreover, you argued as recently as November 2004 in relation to Freitas' proposal that:

"This all looks doable today. It's only a fraction of the way to diamondoid molecular manufacturing. But it's a very important fraction. If it works, it will demonstrate once and for all that mechanically guided diamond-building vacuum chemistry is feasible. The recipe is detailed enough that it's already hard to argue it can't work. The claim that diamond-building can't be used to build machine parts seems likely to lose most of its remaining credibility." (Taken from Diamond building proposal on this blog).

You might now see diamondoid as a 'very narrow proposal' - that's fine, you're more than entitled to revise your opinion. Just don't pretend that the focus on diamondoid is an issue associated with the scientific community's perception/interpretation of Drexler's proposals. (This type of faux 'straw man' argument has been raised previously - see Is Mechanosynthesis Feasible? ). Drexler spends a *considerable* amount of time in "Nanosystems" discussing diamondoid (as we've explored previously, The Mechanosynthesis Debate) and in Chapter 16 outlines a route to molecular manufacturing which **bootstraps** from a polymer-based assembly mechanism to the final stage of computer-controlled *atomic precision* mechanochemistry.

"If it looks like we're now trying to "move the goalposts" and include too much, that may be because the definition has always been that broad, but people have until recently been focusing on a very narrow proposal (i.e. diamond built in vacuum). Don't forget that protein engineering was the first proposed avenue toward molecular manufaturing--proposed in a PNAS paper in 1981 by Drexler" - from the preceding post.

To reiterate: the focus is on diamondoid because Drexler spends a substantial fraction of "Nanosystems" discussing diamondoid. Moreover, if we are to stick to the vision outlined in Nanosystems , Engines of Creation , and Unbounding The Future , molecular manufacturing necessitates computer-controlled molecule-by-molecule (or atom-by-atom) chemistry. This is *not* a very broad definition and is a simple concept to grasp.

Best wishes,

Philip

[Philip Moriarty]

Chris,

"Nebulous terminology and lack of specificity is what has allowed the NNI to think it's funding lots of work that's relevant to molecular manufacturing..."

Does the NNI really believe that it's funding work relevant to molecular manufacturing (in the Drexler sense)? Was this something that came up at the NAS committee meeting? I'd appreciate it if you could point me to a website (or some other evidence) that supports your conclusion that the NNI believes it's supporting working relevant to molecular manufacturing.

Thanks and best wishes,

Philip

[Chris Phoenix, CRN]

OK, Philip, I've found the URL you posted. Here's a clickable version of the link you gave, to Howard Lovy's blog.

I didn't realize you were worried about us trying to move the goalposts *that* far! I don't think anyone claims that Eigler's spelling "IBM" was a demo of molecular manufacturing. Certainly the MM crowd doesn't want MM to be defined that broadly--as shown by the fact that we complain when our NNI claims to be working toward the goals of MM while actually spending most of their energy on other stuff.

There's a difference between claiming something as MM, and citing it as an answer to an objection or a demonstration of an enabling technology. Lots of stuff has been cited for those purposes.

Chris

[Philip Moriarty]

Chris,

Let's not start talking 'past' each other again! Eigler's atomic/molecular manipulation work (and related work by other groups, e.g. Obayu et al.) has been cited as an 'existence proof' for mechanosynthesis. I am well aware that there is a strong difference between the terms "molecular manufacturing" and "mechanosynthesis". Nevertheless, molecular manufacturing - in it's Drexlerian sense - derives directly from mechanosynthesis. If molecular manufacturing, as you see it, no longer requires atomic precision molecular manipulation under computer control then I ask again that CRN *clearly* modifies its message.

[In addition, and as we seem to be communicating again, there are a number of threads in the debate (and on Soft Machines, e.g. Is mechanosynthesis feasible? ) that you have yet to address. For one thing, I'm interested in hearing your comments re. effectively error-free "digital" chemistry in mechanosynthesis. If you'd like to respond on this point it is probably best if you post the comment on Soft Machines under the thread cited above].

Best wishes,

Philip

[Chris Phoenix, CRN]

The "molecular structures" in my definition was meant to imply "atomically precise." That is definitely part of the definition of MM and I certainly am not trying to argue it away!

In Nanosystems, Drexler states that diamond is interesting to analyze because it's one of the harder systems (implying that if you can do that, you can probably also do easier ones). I'm not saying Drexler didn't put diamond at the center of his book. But if you took all the diamond out of Nanosystems, you'd be left with half the book unchanged, and another quarter just needing new material performance numbers and feature sizes for the analysis.

Diamond is the most studied proposal. But a system that built cubic boron nitride in vacuum, or silica in water, using atomically precise methods, would obviously count as a huge success. A system that built programmable polymers in water using nanoscale programmable machines to build and/or position them into precise structures would be also be an exciting example of MM, though its practical use would depend on the polymer and actuators being used and the tools that could be built from them.

I talked with Seeman briefly about his machine. He told me, "We're just doing chemistry." In my presentation to the NAS, I said that I thought his work was in a gray area near mechanosynthesis, and I thought that it might not be too hard to extend it to something that would definitely count as mechanosynthesis. I asked him afterward if I had said anything that made him uncomfortable, and he said no, that was fine.

As to the NNI: I'm not sure which of the people there want MM to go away, which of them think they're already working on it, and which want to kill it but pretend that they're working toward the same goals. But in conversation with one of the people who run it, he made it clear that he thought the NNI was developing tools that would eventually be useful for MM and related goals. Not that NNI was working directly toward MM--I don't think anyone says that. But, yes, working to build a toolbox that will eventually be useful for atomically precise control of matter (without reference to whether it uses MM's methods).

On Eigler, can you point me at the citation you're talking about? There's a difference between "A proves that objection X to idea B is invalid," "A proves that idea B can work," and "A is an example of B." The first is a valid use of Eigler, the second is perhaps sloppy, the third is likely improper. Without seeing the citation I can't comment on which it was.

Chris

[Chris Phoenix, CRN]

Philip, I just read through the discussion you linked (all 9,700 words of it), then did a text search for "digital" and "error." I think I've answered your questions (posts 39 and 40); if not, please ask again.

[Philip Moriarty]

Chris,

"On Eigler, can you point me at the citation you're talking about? "

How about, for example:

"The principles of molecular nanotechnology are being demonstrated daily in government and industry laboratories world wide, including the arrangement of 35 xenon atoms to spell out "IBM" [Eigler]....." [http://www.islandone.org/MMSG/NSSNanoPosition.html ]

This is from a paper by Tihamer Toth-Fejel (where he argues that "molecular nanotechnology...is the culmination of many fields" where "each of these fields reaches its ultimate in precise, molecular control, which is the ability to build large structures to complex, atomic specifications by direct positional selection of reaction sites [Drexler]".(See also quotes on universal assemblers in this paper and cf my recent post on Soft Machines).

Philip